The Comparison of Segmentation Results for UAV High-Resolution Remote Sensing Image among eCognition, ENVI and EDISON

The image segmentation is the basis of recognition and classification in digital image field. Choosing appropriate segmentation methods and software is necessary. The object-oriented software, eCognition and ENVI, and the software of EDISON based on Mean Shift algorithm, any of which can be used to segment image. This research segmented the pretreated UAV high-resolution sensing image by these software and compared the segmentation results. The result shows: The segmentation results effect of the same type of ground objects with complex texture by eCognition is superior to that of ENVI and EDISON. The segmentation effect of the same type of ground objects with similar spectral character by ENVI and EDISON is superior to that of eCognition.